Re: kernel-2.4.9-13SGI_XFS_PR1 bombs

[Seth Mos <knuffie@xxxxxxxxx>]

At 15:04 8-11-2001 -0600, Dan Yocum wrote:
> Hi guys,
>
> I know the Red Hat kernel-2.4.9-13SGI_XFS_PR1 is in testing still, but
> massive I/O errors occur when I run bonnie++ on my 1.12TB FS - the rest of
> the partitions seem to be behaving themselves, tho.

I am not seeing it here even after pounding on it heavily. I do have 
problems that performance degrades over time (> 1 day). But I think this is 
more of a VM issue.

The good news is that this is still 30x faster then the old NCR MP-RAS box :-)
My manager was used to seeing performance halved when starting a second 
database session. Now this happens after the 3rd concurrent session. Even 
when running 8 concurrent dump and reload session in Progress 8 and 9 
databases it handles it well.

Oh yeah, before I forget. People with AMI megaraid controllers should 
update the firmware of their card. The newer version is a lot faster. Going 
from 1.57/3.13 to 1.61j/3.17 improves the read perfomance in the same 
configuration (raid 1 or 10). In my case it went from 58MB/sec to 71MB/sec 
which is a nice perfomance upgrade without costing anuything :-)

> I'll try to get the error messages next time I boot it up.

Yes please. This might be related to the size of your fs.

Cheers

--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.